OpenAI · 2026-04-21 · major
GPT Image 2 — OpenAI's Reasoning-Augmented Image Model with 2K Resolution
OpenAI shipped GPT Image 2 powering ChatGPT Images 2.0 — a model that reasons before generating, supports up to 8 consistent images per prompt, produces up to 2K resolution, and substantially improves text rendering including non-Latin scripts. Available via API as gpt-image-2.
OpenAI's new image model thinks before it generates — reasoning and optionally searching the web before producing images.
Key specs
| Max resolution | 2048px |
|---|---|
| Max batch | 8 images per prompt |
| Model id | gpt-image-2 |
What is it?
GPT Image 2 is OpenAI's latest image generation model, powering the ChatGPT Images 2.0 update. It adds an optional thinking mode that reasons through the prompt and can search the web before generating, similar to how o-series models reason through text tasks. It supports output up to 2K resolution, aspect ratios from 3:1 to 1:3, and can produce up to 8 consistent images from a single prompt — maintaining character and style across all of them.
How does it work?
The model runs in two modes: Instant (fast, direct generation) and Thinking (planning pass before generation, optionally with web search). A dedicated text-rendering subsystem improves handling of small text, dense UI elements, iconography, and non-Latin scripts (Japanese, Korean, Hindi, Bengali). The Thinking mode spends compute up front understanding the intent, which helps with complex compositions, accurate diagrams, and layout-sensitive design work.
Why does it matter?
Prior image models struggled with text fidelity, complex layouts, and consistent multi-image outputs. GPT Image 2's Thinking mode addresses all three directly. The 8-image consistent-batch feature opens up storyboards, character sheets, and multi-panel marketing assets. For developers, the API supports Batch API compatibility with a 50% discount and token-based pricing rather than flat per-image fees.
Who is it for?
Developers building image generation into products; designers needing text-accurate or complex-layout images.
Try it
model='gpt-image-2' in /v1/images/generations or /v1/images/edits